An operating performance assessment strategy with multiple modes based on least squares support vector machines for drilling process. (September 2021)

• Record Type: Journal Article
• Authors: Fan, Haipeng; Wu, Min; Cao, Weihua; Lai, Xuzhi; Chen, Luefeng; Lu, Chengda; Du, Sheng; She, Jinhua
• Is Part Of: Computers & industrial engineering. Volume 159(2021)
• Publication Date: 2021-09
• Subjects: Drilling process -- Multiple modes -- Operating performance assessment -- Process capability index -- Support vector machine; Engineering -- Data processing -- Periodicals; Industrial engineering -- Periodicals; 620.00285
• DOI: 10.1016/j.cie.2021.107492
• Languages: English
• ISSNs: 0360-8352

Abstract

Highlights: A novel operating performance grade classification model with the Taguchi capacity index is proposed. Multiple modes-based operating performance assessment strategy is developed. The proposed method shows better assessment accuracy in actual fields. Abstract: Rate of penetration (ROP) is one of the crucial drilling condition monitoring parameters due to its vital role in real-time assessing drilling operating performance. Operators often adjust operating parameters to meet higher performance requirements. Therefore, drilling operating performance assessment is critical for controlling and optimizing of the drilling process. An operating performance assessment strategy with multiple modes based on least square support vector machines for drilling process is presented in this paper. First, the process capability index is to be taken as the indicator of the ROP and defines the drilling operating performance. Next, the K -means clustering algorithm is used to identify the operating modes. Then, for each mode, an individual drilling operating performance assessment model is established by the method of least squares support vector machines. Finally, drilling operating performance grade is obtained, and actual data of drill well are used for experiments. Further comparative analyses were performed with other state-of-the-art methods, including the Decision tree, Support vector machines (SVM), Least squares support vector machines (LS-SVM), Principal component analysis (PCA), and Partial least squares (PLS). Simulations revealed that the proposed method results in the accurate assessment of operating performance in the drilling process with the accuracy of 87%, the precision of 85.3%, the recall of 88.2%, and the F -Score of 87.6%. In particular, the assessment accuracy was improved by 18.6%, 11.3%, 5.2%, 9.68%, 8.32% in comparison to Decision Tree, SVM, LS-SVM, PCA, and PLS. Performance comparisons reflect the superiority of our model that can ensure high accuracy about operating performance in a drilling process.